{
  "architectures": [
    "LlamaForCausalLM"
  ],
  "model_type": "llama",
  "vocab_size": 50000,
  "hidden_size": 1024,
  "intermediate_size": 3072,
  "num_hidden_layers": 26,
  "num_attention_heads": 16,
  "num_key_value_heads": 8,
  "rms_norm_eps": 1e-06,
  "rope_theta": 500000.0,
  "max_position_embeddings": 4096,
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16"
}